康复机器人治疗的脑卒中患者健侧代偿机制的动态演进建模及在康复评估中的应用

Rehabilitation robot can collect physiological data while providing training aid to the stroke patients. By reflecting in-time rehabilitation effects, it provides a new method to accurately and timely monitor the training process and thus possibly accordingly adjust the treatment. The traditional assessment methods may not be suitable in the context of multi-source, massive data collected by rehabilitation robot. To strengthen the advantages in data collecting, this study is to explore the evolution pattern of sound limbs during robotic rehabilitation. It will analyze the correlation between compensation mechanism decrease of sound sides and functional increase of affected sides as well as how robotic rehabilitation affects this correlation. Then the study proposes a method to recognize and fuse the signals featuring sound limbs compensation. Then a sound limbs compensation evolution model is established as well as its correlation pattern with affected limbs function evolution. Based on that, a rehabilitation assessment method is proposed specific for rehabilitation robots, which may better predict limb coordination after training and analyze no function improvement in affected limbs when trained by certain treatment.

康复机器人在提供训练辅助的同时可以根据采集的数据，实时呈现训练效果，提供了一种及时准确地的监控、调整康复训练方案的新手段。康复机器人采集的数据多源、庞大，传统的定性或小规模定量数据评估的策略很难满足其评估的需求。本研究从发挥康复机器人在生机信号/数据采集方面的效率和广度优势出发，研究脑卒中患者在康复机器人辅助训练过程中，健侧代偿机制变化的规律；分析健侧代偿的减弱与患侧功能增强的相关度，以及康复机器人对这种关联度的影响。构建健体代偿机制表征信号辨识和融合方法，建立健侧代偿机制变化的表达模型，构建健侧与患侧功能变化的关联模型，进而生成一种适合康复机器人训练脑卒中患者的康复评估策略。考虑健体代偿机制变化的康复评估策略能更好预测康复训练对肢体协同能力的作用，也为分析训练中患侧功能不增进等问题提供了有效的工具。

实时监测和评估康复机器人引导下的上肢康复训练是保证治疗安全的基本前提，如何从康复机器人采集的多源庞大数据中提取有用的评估数据具有重要的临床意义。本研究以康复训练中需要重点关注的健体代偿为研究对象，构建了一套在康复运动过程中实时识别及评估训练异常的评估工具，支持临床上少人化/无人化的安全精准康复训练。.本项目通过研究康复机器人引导下健体代偿的产生机制，健侧和患侧的运动关联模型，建立基于实时运动感知网络的上肢康复训练异常运动模式的描述模型、评估模型。并开发了基于代偿评估模型的实时康复训练评估平台，通过临床应用验证，论证考虑健体代偿机制的康复评估分析和康复训练调整的有效性。.本项目在理论上为临床研究康复运动的代偿机制提供了数据依据；项目所搭建的基于视觉的人体运动的实时感知及分析系统，在满足研究需求的同时，也为临床上推广康复评估提供了易操作、低成本的平台，为后续从康复医学、行为感知领域提供了有力的工程佐证。